1 ## dist.topo.R (2010-01-25)
3 ## Topological Distances, Tree Bipartitions,
4 ## Consensus Trees, and Bootstrapping Phylogenies
6 ## Copyright 2005-2010 Emmanuel Paradis
8 ## This file is part of the R-package `ape'.
9 ## See the file ../COPYING for licensing issues.
11 dist.topo <- function(x, y, method = "PH85")
13 if (method == "score" && (is.null(x$edge.length) || is.null(y$edge.length)))
14 stop("trees must have branch lengths for Billera et al.'s distance.")
15 nx <- length(x$tip.label)
18 bp1 <- .Call("bipartition", x$edge, nx, x$Nnode, PACKAGE = "ape")
19 bp1 <- lapply(bp1, function(xx) sort(x$tip.label[xx]))
20 ny <- length(y$tip.label) # fix by Otto Cordero
21 ## fix by Tim Wallstrom:
22 bp2.tmp <- .Call("bipartition", y$edge, ny, y$Nnode, PACKAGE = "ape")
23 bp2 <- lapply(bp2.tmp, function(xx) sort(y$tip.label[xx]))
24 bp2.comp <- lapply(bp2.tmp, function(xx) setdiff(1:ny, xx))
25 bp2.comp <- lapply(bp2.comp, function(xx) sort(y$tip.label[xx]))
29 if (method == "PH85") {
33 if (identical(bp1[[i]], bp2[[j]]) | identical(bp1[[i]], bp2.comp[[j]])) {
39 dT <- q1 + q2 - 2 * p # same than:
40 ##dT <- if (q1 == q2) 2*(q1 - p) else 2*(min(q1, q2) - p) + abs(q1 - q2)
42 if (method == "score") {
49 if (identical(bp1[[i]], bp2[[j]]) | identical(bp1[[i]], bp2.comp[[j]])) {
50 if (i == 19) browser()
51 dT <- dT + (x$edge.length[which(x$edge[, 2] == nx + i)] -
52 y$edge.length[which(y$edge[, 2] == ny + j)])^2
53 found1 <- found2[j] <- TRUE
57 if (found1) found1 <- FALSE
58 else dT <- dT + (x$edge.length[which(x$edge[, 2] == nx + i)])^2
61 dT <- dT + sum((y$edge.length[y$edge[, 2] %in% (ny + which(!found2))])^2)
67 .compressTipLabel <- function(x)
69 ## 'x' is a list of objects of class "phylo" possibly with no class
70 if (!is.null(attr(x, "TipLabel"))) return(x)
71 ref <- x[[1]]$tip.label
72 if (any(table(ref) != 1))
73 stop("some tip labels are duplicated in tree no. 1")
75 for (i in 2:length(x)) {
76 if (identical(x[[i]]$tip.label, ref)) next
77 ilab <- match(x[[i]]$tip.label, ref)
78 ## can use tabulate here because 'ilab' contains integers
79 if (any(tabulate(ilab) > 1))
80 stop(paste("some tip labels are duplicated in tree no.", i))
82 stop(paste("tree no.", i, "has different tip labels"))
83 ie <- match(1:n, x[[i]]$edge[, 2])
84 x[[i]]$edge[ie, 2] <- ilab
85 x[[i]]$tip.label <- NULL
87 x[[1]]$tip.label <- NULL
88 attr(x, "TipLabel") <- ref
92 prop.part <- function(..., check.labels = TRUE)
95 if (length(obj) == 1 && class(obj[[1]]) != "phylo")
98 ## class(obj) <- NULL # needed?
101 if (ntree == 1) check.labels <- FALSE
102 if (check.labels) obj <- .compressTipLabel(obj)
103 for (i in 1:ntree) storage.mode(obj[[i]]$Nnode) <- "integer"
105 ## The 1st must have tip labels
106 ## Maybe simply pass the number of tips to the C code??
107 if (!is.null(attr(obj, "TipLabel")))
108 for (i in 1:ntree) obj[[i]]$tip.label <- attr(obj, "TipLabel")
110 clades <- .Call("prop_part", obj, ntree, TRUE, PACKAGE = "ape")
111 attr(clades, "number") <- attr(clades, "number")[1:length(clades)]
112 attr(clades, "labels") <- obj[[1]]$tip.label
113 class(clades) <- "prop.part"
117 print.prop.part <- function(x, ...)
119 if (is.null(attr(x, "labels"))) {
120 for (i in 1:length(x)) {
121 cat("==>", attr(x, "number")[i], "time(s):")
122 print(x[[i]], quote = FALSE)
125 for (i in 1:length(attr(x, "labels")))
126 cat(i, ": ", attr(x, "labels")[i], "\n", sep = "")
128 for (i in 1:length(x)) {
129 cat("==>", attr(x, "number")[i], "time(s):")
130 print(x[[i]], quote = FALSE)
135 summary.prop.part <- function(object, ...) attr(object, "number")
137 plot.prop.part <- function(x, barcol = "blue", leftmar = 4, ...)
139 if (is.null(attr(x, "labels")))
140 stop("cannot plot this partition object; see ?prop.part for details.")
142 n <- length(attr(x, "labels"))
143 layout(matrix(1:2, 2, 1), heights = c(1, 3))
144 par(mar = c(0.1, leftmar, 0.1, 0.1))
145 plot(1:L, attr(x, "number"), type = "h", col = barcol, xlim = c(1, L),
146 xlab = "", ylab = "Frequency", xaxt = "n", bty = "n")
147 plot(0, type = "n", xlim = c(1, L), ylim = c(1, n),
148 xlab = "", ylab = "", xaxt = "n", yaxt = "n")
149 for (i in 1:L) points(rep(i, length(x[[i]])), x[[i]], ...)
150 mtext(attr(x, "labels"), side = 2, at = 1:n, las = 1)
153 prop.clades <- function(phy, ..., part = NULL)
157 if (length(obj) == 1 && class(obj[[1]]) != "phylo")
158 obj <- unlist(obj, recursive = FALSE)
159 part <- prop.part(obj, check.labels = TRUE)
161 bp <- .Call("bipartition", phy$edge, length(phy$tip.label),
162 phy$Nnode, PACKAGE = "ape")
163 if (!is.null(attr(part, "labels")))
164 for (i in 1:length(part))
165 part[[i]] <- sort(attr(part, "labels")[part[[i]]])
166 bp <- lapply(bp, function(xx) sort(phy$tip.label[xx]))
167 n <- numeric(phy$Nnode)
168 for (i in 1:phy$Nnode) {
169 for (j in 1:length(part)) {
170 if (identical(all.equal(bp[[i]], part[[j]]), TRUE)) {
171 n[i] <- attr(part, "number")[j]
180 boot.phylo <- function(phy, x, FUN, B = 100, block = 1, trees = FALSE)
182 if (is.list(x) && !is.data.frame(x)) {
183 if (inherits(x, "DNAbin")) x <- as.matrix(x)
189 x <- matrix(x, n, nL/n, byrow = TRUE)
193 boot.tree <- vector("list", B)
196 y <- seq(block, ncol(x), block)
197 boot.i <- sample(y, replace = TRUE)
198 boot.samp <- numeric(ncol(x))
199 boot.samp[y] <- boot.i
200 for (j in 1:(block - 1))
201 boot.samp[y - j] <- boot.i - j
202 } else boot.samp <- sample(ncol(x), replace = TRUE)
203 boot.tree[[i]] <- FUN(x[, boot.samp])
205 for (i in 1:B) storage.mode(boot.tree[[i]]$Nnode) <- "integer"
206 storage.mode(phy$Nnode) <- "integer"
207 ans <- attr(.Call("prop_part", c(list(phy), boot.tree),
208 B + 1, FALSE, PACKAGE = "ape"), "number") - 1
209 if (trees) ans <- list(BP = ans, trees = boot.tree)
213 consensus <- function(..., p = 1, check.labels = TRUE)
215 foo <- function(ic, node) {
217 ## node: node number in the final tree
220 for (j in (ic + 1):m) {
221 wh <- match(pp[[j]], pool)
222 if (!any(is.na(wh))) {
223 edge[pos, 1] <<- node
225 edge[pos, 2] <<- nextnode <<- nextnode + 1L
233 ind <- pos:(pos + size - 1)
234 edge[ind, 1] <<- node
235 edge[ind, 2] <<- pool
240 if (length(obj) == 1) {
241 ## better than unlist(obj, recursive = FALSE)
242 ## because "[[" keeps the class of 'obj':
244 if (class(obj) == "phylo") return(obj)
246 if (!is.null(attr(obj, "TipLabel")))
247 labels <- attr(obj, "TipLabel")
249 labels <- obj[[1]]$tip.label
250 if (check.labels) obj <- .compressTipLabel(obj)
253 ## Get all observed partitions and their frequencies:
254 pp <- prop.part(obj, check.labels = FALSE)
255 ## Drop the partitions whose frequency is less than 'p':
256 pp <- pp[attr(pp, "number") >= p * ntree]
257 ## Get the order of the remaining partitions by decreasing size:
258 ind <- sort(unlist(lapply(pp, length)), decreasing = TRUE,
259 index.return = TRUE)$ix
263 edge <- matrix(0L, n + m - 1, 2)
272 structure(list(edge = edge, tip.label = labels,
273 Nnode = m), class = "phylo")